VPG: Visual Prefix Guidance for Autoregressive Image and Video Generation 事件

VPG: Visual Prefix Guidance for Autoregressive Image and Video Generation arXiv:2605.30317v1 Announce Type: new Abstract: Autoregressive image and video generators are trained with teacher-forced histories but must sample from their own generated prefixes at inference time, making them vulnerable to exposure bias and prefix drift. Existing remedies either modify training or apply sampling-time guidance aimed primarily at external semantic conditions, such as class labels or text prompts, rather